God’s Eye View: a globe, AI interaction and honest data boundaries
Measure God’s Eye View honestly: startup, source load, labels and optional provider use
Interpret the repository’s hardware-specific baseline and design your own measurements without promising universal frame rates or free operation.
What you will learn
- The published baseline is a dated observation
- More labels can cost more than more controls
- Keyless and quota-aware do not mean unlimited
Before you start
- Basic JavaScript and JSON
- Understanding of coordinates and source timestamps
Explain how source records, validation, layer presentation and optional AI interaction differ, using synthetic examples.
Key takeaways
- A reported hardware baseline is not a universal promise.
- Cold loading, labels and heap deserve separate measurements.
- Application throttles are not provider billing caps.
The published baseline is a dated observation
The performance document records an Apple M5, Chrome 150, 1440×900 capture from August 2026 and explicitly says the original artifacts are not included. It is a reported comparison, not a runnable benchmark or minimum hardware specification. We did not reproduce it.
Separate application-ready from initial visual settling, cold layer activation from warm option changes, and moving from resting frames. Network-dependent source population matters as much as GPU settings. A scene with no live rows cannot fairly stand in for a dense populated scene.
More labels can cost more than more controls
The document’s overlay and stress sections show why text draws, observation counts, selected labels and heap are useful measures alongside frame rate. A display ceiling can hide differences between light scenes, while source loading and memory still vary substantially.
For your own controlled test, hold viewport, device pixel ratio, camera path and synthetic record count constant. Record long tasks and activation delay before adding another visual effect. The fixed 96 earthquake-label cohort is a local presentation limit, not a guarantee that the entire app stays within a chosen memory budget.
Keyless and quota-aware do not mean unlimited
Map tiles, voice and other optional providers have separate accounts and usage rules. Application caches and request budgets reduce repeated work but do not replace provider-side quotas. Per-IP in-memory throttles reset with the process and are not billing caps.
Measure actual provider usage on an authorized test account before estimating costs. Keep manual keyless exploration as a separate baseline from a voice-enabled or photorealistic scene. This article reports no new FPS, startup time, voice price or cost-saving percentage.
Decision guide
| Criterion | Option A | Option B |
|---|---|---|
| Best when | You need predictable behavior and easy auditing | You need adaptive optimization and have reliable telemetry |
| Main risk | May leave performance on the table | Can become difficult to explain or debug |
Implementation steps
- 1
Fix hardware, viewport and scene population.
- 2
Measure ready, settle, activation and frame behavior separately.
- 3
Record labels, heap and provider requests.
- 4
Verify current quotas before enabling paid capabilities.
Copy-ready example
{
"measurementPlan": true,
"viewport": [
1440,
900
],
"devicePixelRatio": 1,
"syntheticRecords": 100,
"readyMs": null,
"settleMs": null,
"heapMiB": null,
"frameRate": null,
"providerUsage": null,
"benchmarkExecuted": false
}Frequently asked questions
Can I expect the README’s startup number on my machine?
No. The underlying document gives a specific hardware and browser context.
Do per-IP limits cap the monthly bill?
No. The security guide describes process-local throttles, not billing caps.
Sources
- God’s Eye View / docs/PERFORMANCE.mdSource checked 2026-09-14
- God’s Eye View / SECURITY.mdSource checked 2026-09-14
- God’s Eye View / src/layers/earthquakes/model.jsSource checked 2026-09-14
- God’s Eye View / README.mdSource checked 2026-09-14